Intergenerational effects of endocrine-disrupting compounds: a review of the Michigan polybrominated biphenyl registry.

Flame retardants have been used in electronic appliances, textiles, plastics and polyurethane foams since the increase in fire safety legislation in the 1960s, with one of the most common flame retardants being PBB. From 1970 to 1976 alone, 13.3 million pounds of PBBs were produced in the USA. Over 88% of all PBBs were produced by Velsicol's Michigan Chemical Company as FireMaster BP-6 and FireMaster FF-1 [ 6 ]. In May 1973, a shipment of FireMaster FF-1 was mistakenly shipped to Farm Bureau Services feed mill instead of NutriMaster (magnesium oxide), a nutritional supplement for livestock feed. Both substances were produced by the Michigan Chemical Company, both were fine white powders, and both were stored in plain bags with black lettering [ 7 ]. It is not known how much FireMaster FF-1 was sent to the Farm Bureau, but the estimates range from 500 to 1000 pounds [ 8 ]. In July 1973, the Farm Bureau mixed the FireMaster FF-1 into their animal feed, and during that fall, contaminated animal feed with high levels of PBBs (4000–13,500 ppm) was shipped to farms around Michigan [ 7 , 8 ]. The contaminated feed was no longer on the market by December 1973, but animals were fed with it until May 1974, when PBB was identified as the cause of the toxic effects seen in the farm animals [ 7 , 8 ]. Additionally, lower levels of contamination still occurred from several sources. Mill machinery still contained PBB residues and continued to contaminate new animal feed. One mill alone contaminated over 3000 pounds of feed in this manner. Feed swapping between mills and recycling contaminated animals products for new animal feed added to the number of mills and animals that were contaminated [ 8 ]. This caused a lower level of contamination, but it continued until the mills were cleaned a year later [ 7 , 8 ]. After the introduction of feed contaminated with high concentrations of PBB at dairy farms across Michigan, cows’ feed consumption decreased by 50%, followed by a decrease in milk production. Cows also had difficulties reproducing and calving, had abnormal hoof growths, hyperkeratosis and hematomas. Abscesses were found on the dead cows’ livers and other organs, as were intra-abdominal adhesions [ 9 , 10 ]. Cattle were underweight, had fibroids in their lungs, enlarged thymus glands and lesions and lymphocyte infiltration in their livers and kidneys. The bulls had a reduced number or complete lack of mature spermatozoa and atrophied testes. Furthermore, the small intestine showed lymphocyte infiltration, which may explain the decreased feed consumption [ 10 ]. With the cause of the health problems in the livestock finally identified as the contaminated feed, efforts focused on stopping the contamination from spreading. The US FDA regulated that PBB levels must be less than 1 ppm in dairy products intended for human consumption. The Michigan Department of Agriculture identified and quarantined contaminated farms. Although the quarantine started with dairy farms, other livestock had to be quarantined because of the cross contamination at the Farm Bureau mills. Over 30,000 cattle, 2000 pigs, 400 sheep and 2 million chickens had to be destroyed and buried in state-approved burial sites [ 7 ]. Given the toxic effects observed in exposed animals, human exposure to PBBs was a concern. It was determined that approximately 6.5 million people in Michigan had some PBB exposure, with higher exposures among those who lived on quarantined farms or worked at the Michigan Chemical Company [ 7 , 8 , 11 ]. In the mid-1970s, approximately 5000 individuals consented to participate in a study of the long-term effects of PBBs on humans ( Figure 1 ). Many of the registrants’ children have since been enrolled [ 8 ].

The Michigan PBB registry is an active cohort and the community is involved in investigating their continuing health risks. Despite the predominant exposure having occurred over 40 years ago, PBBs are highly lipophilic and bioaccumulative, meaning that exposure and its consequences to the population continues long after the resolution of the initial contamination event [ 7 ]. PBBs are able to pass through the placental barrier and are present in breast milk and are thus able to expose future generations. Therefore, people are still at risk from PBB exposure. While studies in animal models are highly informative for guiding epidemiological investigations, it is difficult to translate knowledge from early studies of animal models to humans because of the differences in quantifying PBB exposure between the two experimental designs. Specifically, most PBB studies in animals report the amount of PBB added to the animal's diet instead of measuring PBB levels in blood and other tissues assayed in humans. Animal studies also rarely assess exposure levels in the offspring of exposed animals. As PBB exposure remains an active public health issue in Michigan, additional studies of animal models exposed to PBBs will provide more precise information regarding the effects of PBB exposure and additional insight into the mechanisms underlying intergeneration transmission of health consequences from exposure. To enable a better understanding of the health risks to those exposed and subsequent generations, epigenetic and genetic data should be collected in longitudinal cohorts like the Michigan PBB registry. The addition of these datasets to well-studied epidemiological cohorts will allow future studies to better investigate the effects of EDCs on human health. Identification of genes whose function differ in response to PBB exposure will increase understanding of how PBB contributes to health problems. Because the Michigan PBB registry has been followed for 40 years and is adding subsequent generations of those originally exposed, investigating epigenetic and genetic associations will lead to a greater understanding of the biological mechanisms behind the intergenerational transmission of exposure-related health effects and help to determine the health risks for future generations. Results from this cohort can be used to help guide studies on other types of EDCs and exposures and to determine if there is a safe exposure level to EDCs for humans. Endocrine-disrupting compounds (EDCs) are a growing concern for the modern world, and multigenerational cohorts are needed to study it further. The Michigan polybrominated biphenyl registry is a multigenerational cohort who were exposed to an EDC 40 years ago during an agricultural accident and have been followed since. Existing studies in this cohort show an intergenerational transmission of exposure-related health problems. Genetic and epigenetic studies of health problems in the Michigan polybrominated biphenyl registry may lead to a greater understanding of the mechanisms behind intergenerational transmission of EDCs and the risk to future generations.

A cohort of approximately 5000 exposed individuals from Michigan was established soon after the contamination incident and has been used in many studies evaluating the effects that PBBs. The first studies with this cohort began in the late 1970s. On a questionnaire asking about the incidence of health problems, ranging from acne to heart disease, the parents of 292 children who were exposed to PBBs reported a higher prevalence of symptoms in their children than parents of unexposed children from Wisconsin. However, when serum PBB levels were analyzed in these children, it was discovered that the highest number of symptoms were reported in the children who had the lowest serum PBB levels, and that the only complaint that had a positive association with PBB levels was acne in 10–11 year olds [ 31 ]. A later study done in both adults and children from the quarantined and nonquarantined farms, as well as the Michigan Chemical Company workers, showed no increase in symptoms in the children from the farms compared with control children in Wisconsin. However, the study did show that adults exposed to PBBs have reported significantly more alopecia, halogen acne, skin peeling and scaling, erythema, itching, increased finger nail growth and increased sweating. It is unclear why adults would experience skin problems when exposed to PBBs and children would not, especially since the children and farmers were both exposed in the same manner [ 32 ]. People in Michigan were concerned with the persistence of PBBs in their bodies. As discussed above, the lipophilic properties PBBs make it likely to persist in humans for years. A study done in 406 women from the cohort determined that the PBB elimination rates in each person were modified by a variety of factors. Higher BMI at enrollment was associated with a slower rate of elimination, and smoking was associated with a faster rate of elimination. The elimination rate was significantly faster if a woman had been pregnant and breastfeeding between serum measurements versus if she had just been pregnant (a decrease of 0.007 ppb PBB/year if pregnant and breastfeeding vs an increase of 0.008 ppb PBB/year if just pregnant), although there was not a significant difference compared with women who had never been pregnant. This was surprising given that breastfeeding in animals was shown to be a major elimination route for PBBs but could be due to the mobilization of PBBs from adipose tissue to the bloodstream during physiological changes like pregnancy [ 33 ]. Using elimination rates based on these factors, the biological half-life for PBBs in humans was calculated to be between 13 and 29 years, depending on the person's BMI and their initial PBB exposure level [ 34 ]. Women who were exposed as children had faster rates of PBB elimination, although that may be related to children having higher PBB serum initially and then undergoing a growth spurt. Additionally, 12% of the women surveyed showed an increase in PBB serum levels, which remains largely unexplained, unless those subjects somehow had continued exposure even after the contamination clean-up or there was a measurement error [ 35 ]. All of these studies were limited because height and weight were only collected at enrollment, not throughout, so any changes in serum PBB levels due to weight changes or growth spurts in children were lost. Furthermore, these studies used only women, making it impossible to assess sex-specific elimination rates and most of the women only had their PBB levels measured twice, making accurate elimination curves difficult. Even with current measurements of PBB exposure, studies of thyroid problems gave conflicting results. Initial reports in 35 chemical plant workers showed that 11% were diagnosed with hypothyroidism and that they had significantly higher levels of thyrotropin, a hormone that regulates the production of other thyroid hormones. However, serum PBB levels were never measured and PBB exposure was assumed based on their employment by the chemical factory, thus the increased prevalence of thyroid problems was unable to be associated with an increased PBB exposure [ 36 ]. These results were disputed by a short report in 51 chemical workers, who had an average PBB level of 12.82 ppm, which is substantially more than the average exposure in Michigan. None of these workers reported hypothyroidism or thyroid enlargement, but the report was not explicit with its methods, making it difficult to determine why these two studies conflicted [ 37 ]. Further studies in 200 men and women tried to correlate reported thyroid problems with PBB exposure, but again there was no difference in thyroid disease incidence with levels of PBB exposure [ 38 ]. These studies may not have had sufficient sample sizes given the large interpersonal variations in thyroid hormone levels. In the most recent study of 753 individuals from the Michigan PBB registry, it was found that 18% of participants (and 25% of women) had thyroid disease, with women exposed to higher levels of PBB having increased odds of developing thyroid disease (odds ratio [OR]: 1.12 for any thyroid disease; OR: 2.13 for hypothyroidism). In participants that did not report thyroid disease, there was no association between thyroid hormone levels and PBB exposure levels [ 39 ]. A study of cancer risk in individuals exposed to higher levels of PBBs found that, for lymphomas and digestive system cancers, which were grouped to include liver, stomach, esophagus and pancreas, the OR for lymphomas for individuals in the highest exposure group (>50 ppb) versus the lowest exposure group (<3 ppb) was 48.9. For digestive cancers, the OR between exposure groups was 22.9, suggesting that PBB exposure does increase the risk for certain types of cancer [ 40 ]. Other studies found that there was an increased risk of breast cancer in people exposed to higher levels of PBB, but it did not reach statistical significance [ 41 , 42 ]. Recently, PBBs have been reclassified from a possible (Group 2B) to a probable (Group 2A) human carcinogen by the International Agency for Research on Cancer. Given that animal studies had shown reproductive delay in rats exposed to PBBs, research was done to determine whether there was an increased incidence of reproductive disorders. Studies conducted to date have not identified an association between PBB exposure and incidence of endometriosis [ 43 ] or risk of spontaneous abortion [ 44 ]. PBB exposure was also not associated with a risk of an abnormal pap test result [ 45 ]. However, these studies were all conducted in individuals that were directly exposed to PBBs. Importantly, the effects of PBBs are not limited to those originally exposed, because PBBs are transmitted to the children of exposed mothers. As in the animal models, PBBs in humans can be passed through the placenta and through breast milk. Umbilical cord blood serum was found to have about a third the PBB concentration of the maternal blood serum, but breast milk was found to have 100-times the concentration of PBBs [ 46 , 47 ]. Additionally, infants breastfed by exposed mothers for 6 months were six-times more likely to have detectable levels of PBBs in their blood than babies who were born to exposed mothers but not breastfed [ 47 ]. Thus, there is substantial evidence of intergenerational transmission of PBBs and that infants born to exposed mothers are exposed during critical developmental periods. Research then assessed the health effects in the second generation. One study reported a higher percentage of boys born to the exposed cohort (54.2%), though this difference was not statistically significant [ 48 ]. Another study found no significant association between either infant birth weight or gestational age and maternal PBB exposure [ 49 ]. However, a further study in this cohort did report that babies born to mothers exposed to high levels of PBBs had higher birth weight (an increase of 16 g per 10 ppb exposure in the mother), but only if the mothers were exposed before puberty [ 50 ]. This was further confirmed by offspring of mothers with even lower PBB exposure (blood serum above 1 ppb) having three-times the odds of having lower Apgar scores, but again, only if the mothers were exposed to prior to the start of puberty (age of exposure was <10 years old) [ 51 ]. Thus, more nuanced approaches to how PBBs affect exposed individuals and their offspring need to be implemented, and the dosage, duration and time point of the exposure need to be considered. Because the Michigan PBB registry has been tracking the exposed cohort for over 40 years now, it has also been possible to study the effects of PBB exposure on adults who were exposed in utero and through breastfeeding. Men born to highly exposed mothers were three-times as likely to report any genitourinary condition, like hernias or hydroceles (OR: 3.1) [ 52 ]. Women born to highly exposed mothers had an earlier age of menarche and an earlier age at pubic hair development (11.6 vs 12.7 years old). While there was no change in the timing of breast development (as measured by the Tanner Stage criteria) in this same study, this may be due to the difficulty of the girls assessing their own breast development based on pictures [ 53 ]. The daughters of highly exposed mothers had a higher OR for spontaneous abortions (OR: 4.08), although they did not have any trouble with conceiving children [ 54 ]. Taken together, this suggests that there may be certain windows of vulnerability during human development, making children of exposed parents more susceptible to adverse health outcomes from the endocrine-disrupting action of PBBs. While the individuals who were directly exposed in the 1970s report a variety of health problems ranging from skin problems to cancer, their children report more reproductive and developmental conditions ( Figure 3 ).

Another way there could be an intergenerational pattern of adverse health effects following chemical exposure would be through families being exposed to both similar environments and transmitting genetic material from parent to child. The health effects following PBB exposure will most likely depend on the genetic architecture of the population and individual exposed. Two people could be exposed to the same chemical, at the same level, in the same manner, at the same age and have different health outcomes due to genetic variation. For example, one study found that the length of the CAG repeat in exon 1 of AR , which varies in the population and the alters the transcriptional activity of AR , mediated the reproductive outcomes associated with exposure to PCB. Men with fewer than 20 CAG repeats who were exposed to PCB above the median exposure level had a 42% decrease in sperm count and a 35% decrease in sperm concentration. Moreover, the authors found that if they did not account for the number of CAG repeats that no association was found between sperm concentration or count and PCB exposure level [ 82 ]. Another study found that both the CAG and the GGN repeat length polymorphisms in exon 1 of the AR gene modified the association between EDCs and the proportion of Y bearing sperm, and this association was more pronounced if the men also had a common SNP (rs2292596) in AHRR that made it less able to repress AHR [ 83 ]. Common variants that mediate the stability of the AHR gene and the repressing function of the AHRR gene also mediate the association between EDCs, including PCBs, and reproductive function [ 84 ]. Differences in genes related to steroid-hormone and xenobiotic pathways may explain why many health outcomes only associate with exposure to EDCs in some populations, and the difficulty that studies have in finding an association between the exposure dose and health outcomes [ 85 , 86 ]. Therefore, the difficulty in associating health outcomes to exposure levels in the Michigan PBB cohort may be resolved by accounting for variation in genetic structure. Additionally, genetic variation can impact epigenetic variation and disease susceptibility. For example, methylation quantitative trait loci are genetic variants where the genotype at that locus results in differing patterns of DNA methylation [ 87 ]. Methylation quantitative trait loci are associated with the development and progression of many human diseases [ 88 , 89 ]. Therefore, some individuals may be more likely to experience epigenetic alterations following EDC exposure. Understanding genetic susceptibility will aid in identifying health risks in parents and their children exposed to EDCs.

While great progress has been made over the past 40 years to determine the health effects of PBB exposure, there are still many research questions that remain unanswered. It is unclear at this time why some of the symptoms and conditions reported more frequently by the individuals in the Michigan cohort do not seem to depend on the concentration of PBBs in their blood, although the difficulty in calculating the appropriate target organ's dose of PBB may contribute to this problem. Additionally, while children of exposed individuals might have more reproductive problems than their parents did due to developmental windows of vulnerability, this hypothesis is not fully tested, and it is not known which effects are sensitive to particular developmental periods. Furthermore, the mechanism by which PBB exposure causes health problems is not fully understood. Changes in epigenetic marks have been linked to the development of many diseases, including those reported by the people in the Michigan PBB registry. Disorders associated with changes in epigenetic marks include many types of cancers, neurological problems, autoimmune disease and reproductive problems [ 55–57 ]. If PBB does influence epigenetic regulation of genes, that could provide a mechanism to explain the increased number of health problems experienced by the people in Michigan. Currently, no research examining epigenetic differences associated with PBB exposure has been reported, either in animal models or in humans. Therefore, hypotheses on how PBB could affect the epigenome have to rely on studies in other EDCs. Because PBBs have a chemical structure that is similar to PCBs and polybrominated diphenyl ethers (PBDEs), PBBs are most comparable with these molecules. Additionally, exposure to PCBs and PBDEs are associated with many of the same health problems as exposure to PBBs, such as increased risk for reproductive problems, issues with steroid hormone metabolism and thyroid problems [ 58–71 ]. Because PBBs, PCBs and PBDEs share the same adverse health outcomes and are structurally similar, studies on how PCBs and PBDEs affect epigenetic marks are currently the best tool to understanding how PBBs affect the epigenome. Research in EDCs show global changes in DNA methylation, histone methylation, gene expression and cellular function. Both human and mouse cell lines that were exposed to micromolar quantities (∼0.6 ppm) of PBDEs and PCBs showed significant decreases in global DNA methylation levels and increases in adipocyte differentiation, potentially explaining the epidemiological studies linking exposure to these EDCs to increased obesity in adulthood [ 72 ]. Additionally, in a human vascular endothelial cell line exposed to 0.3 to 3 nM of PCBs (∼0.09–0.9 ppb), exposure activated demethylation of histone H3 lysine 9 (H3K9), which then caused 140–200% increase in expression of inflammatory genes (vs controls) [ 73 ]. Studies with bisphenol A (BPA), another EDC, showed that breast epithelial cell lines that were exposed to BPA had increased activation of the estrogen signaling pathway, which then led to the suppression of 170 different genes and increased levels of DNA methylation across the genome. The author suggested that the epigenetic changes and increased estrogen activity could explain why exposure to BPA increases the risk of breast cancer development [ 74 ]. Another study reported that EDCs induce estrogen signaling pathways, both through the estrogen receptor's transcription factor activity and through the estrogen receptor's activation of other signaling pathways. Additionally, this study found that EDCs activate the PI3K/AKT pathway, which results in the inactivation of EZH2, a histone methyltransferase. EDCs can also bind the members of the nuclear hormone receptor superfamily, and thus activate the MAPK signaling pathway, which can influence cell growth, apoptosis, differentiation and motility, potentially explaining why exposure to EDCs during early developmental windows disrupts organogenesis and reduces fecundity and sperm motility [ 75 ]. Taken together, EDCs can directly change the epigenetic state of a cell, the transcriptional profile of a cell and that cell's function, which could lead to the health problems seen in exposed populations. In one of the first studies investigating persistent organic pollutants and DNA methylation in humans, PCB exposure was associated with global DNA hypomethylation in a cohort of 71 highly exposed individuals from Greenland [ 76 ]. Another study in 86 Koreans with population-level exposure to PCBs also found global DNA hypomethylation with increasing exposure level [ 77 ]. In a study of 524 elderly people in Sweden, however, increasing PCB exposure was associated with global DNA hypermethylation [ 78 ]. While it is not clear why in some populations EDC exposure is associated with DNA hypermethylation and in others it is associated with DNA hypomethylation, these conflicting results may be due to the different studies using different methods to estimate global DNA methylation levels, differences in the populations, differences in the dose of the exposure or differences in the timing of exposure. Therefore, while exposure to EDCs appears to associate with differences in global DNA methylation, more studies are needed to determine the precise relationship. Additionally, because these global DNA methylation studies only examine methylation in repetitive, gene-poor regions of the genome, genome-wide studies looking at differences in methylation level at CpGs near genes are needed, as well as studies evaluating epigenetics as a mediator between exposure and the health outcomes reported by exposed humans. PCBs have been shown to bind to nuclear receptors and change gene expression, and to associate with differences in DNA methylation levels in human populations that have been exposed. Presumably, PBBs can also bind to nuclear receptors on DNA and drive gene expression, that gene expression may be associated with differences in the epigenetic marks and in cellular function. These differences would be more pronounced with higher exposure levels [ 69 ]. From the animal and human studies, PBBs pass through the placental barrier and are present in breast milk – exposing children born to exposed mothers and leading to health problems in offspring that were not seen in the exposed parents [ 46 , 47 , 51–54 ]. Altered gene regulation and expression early in development could explain why the children of those exposed have more health problems than their parents. Precise epigenetic patterns are vitally important during embryonic development because the specific timing of gene activation and repression determines cell differentiation and organogenesis. Additionally, because epigenetic marks need to change, and to change precisely, for an embryo to develop, these marks may be particularly susceptible to disturbances. Disruption of epigenetic signals by PBBs crossing the placenta could cause improper development or permanently altered marks that would cause life-long problems [ 79 ]. A study done with placentas from the National Children's Study found that PBDEs and PCBs increased DNA methylation in the placenta at imprinted loci. PCB exposure was also negatively associated with H19 expression, which is important for proper regulation fetal growth [ 80 ]. Children who were exposed early in development would be more susceptible to epigenetic alterations due to their epigenetic marks changing and their cell lineages still being determined. This could then lead to an increased risk of developmental and chronic diseases throughout life. Furthermore, while it has not been demonstrated in a human population exposed to EDCs, there have been a report in animal models exposed to EDCs of epigenetic marks escaping mammalian reprogramming events that happen early in development. In this study, the epigenetic marks that associated with environmental exposure were said to be transmitted to offspring even when the offspring was not exposed to the chemical itself [ 81 ]. If this is the case, then the children of exposed men would also be at risk, as well as the children of exposed women.

PBBs are a class of chemicals characterized by two phenyl rings with bromine substitutions. The number and positions of the bromine substitutions yield 209 different configurations of PBB, called congeners. The chemical properties and biological activity of PBB depend on the degree and position of the substitution [ 6 ]. The commercial product in the Michigan incident was FireMaster FF-1, which consisted of another existing technical mixture of PBBs (FireMaster BP-6) and 2% calcium silicate – an anticaking agent [ 7 ]. Twenty-two different congeners of PBBs have been detected in FireMaster BP-6, with the most common congeners being PBB-153 (2,2′,4,4′,5,5′-hexabromobiphenyl; over 50%) and PBB-180 (2,2′,3,4,4′,5,5′-heptabromobiphenyl) [ 6 , 7 ].